SN

TDa

b

c

a

ARRAA

KNS(MT

1

mrdtNcvtTvtdsieNo[

0d

Journal of Alloys and Compounds 482 (2009) 298–301

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

journa l homepage: www.e lsev ier .com/ locate / ja l l com

urface mechanical attrition treatment induced phase transformation behavior iniTi shape memory alloy

. Hua, C.S. Wenb, J. Lub, S.L. Wua, Y.C. Xina, W.J. Zhanga, C.L. Chuc,∗, J.C.Y. Chunga, K.W.K. Yeunga,.T.K. Kwoka, Paul K. Chua,∗∗

Department of Physics and Materials Science, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong, ChinaDepartment of Mechanical Engineering, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong, ChinaSchool of Materials Science and Engineering, Southeast University, 211189 Nanjing, China

r t i c l e i n f o

rticle history:eceived 31 December 2008eceived in revised form 31 March 2009ccepted 1 April 2009

a b s t r a c t

The phase constituents and transformation behavior of the martensite B19′ NiTi shape memory alloy afterundergoing surface mechanical attrition treatment (SMAT) are investigated. SMAT is found to induce theformation of a parent B2 phase from the martensite B19′ in the top surface layer. By removing the surfacelayer-by-layer, X-ray diffraction reveals that the amount of the B2 phase decreases with depth. Differential

vailable online 10 April 2009

eywords:iTi shape memory alloy (SMA)urface mechanical attrition treatmentSMAT)

scanning calorimetry (DSC) further indicates that the deformed martensite in the sub-surface layer up to300 �m deep exhibits the martensite stabilization effect. The graded phase structure and transformationbehavior in the SMATed NiTi specimen can be attributed to the gradient change in strain with depth.

© 2009 Elsevier B.V. All rights reserved.

artensite stabilizationransformation behavior

. Introduction

The near equiatomic NiTi shape memory alloys have attractedore attention over the past two decades due to their supe-

ior properties, e.g. shape memory effect, superelasticity, highamping capacity, excellent biocompatibility and fatigue resis-ance etc. These properties enable the successful applications ofiTi SMAs in various engineering practices [1–4]. Almost all appli-ations of NiTi SMAs involve transformation-related deformationia either martensite reorientation or stress-induced martensiteransformation and shape recovery of deformed martensite [5,6].herefore, the transformation behaviors of the materials underarious deformation conditions are of great concern. To predicthe transformation behavior of NiTi SMAs as a function of variouseformation factors, namely temperature [7], loading mode [8,9],train [10] and grain size [11,12], much work has been conductedn the past few years. Lin and co-workers [8,13] investigated the

ffects of cold-rolling on the transformation behavior of equiatomiciTi SMAs and found the parent B2 phase and stabilization effectsn thermoelastic martensite can be induced by cold-rolling. Liu5,14,15] studied the deformation of NiTi SMAs via martensite reori-

∗ Corresponding author.∗∗ Corresponding author. Tel.: +852 2788 7724; fax: +852 2788 7830.

E-mail addresses: clchu@seu.edu.cn (C.L. Chu), paul.chu@cityu.edu.hk (P.K. Chu).

925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.jallcom.2009.04.004

entation (MR) and considered that the internal elastic energy storedin the reoriented martensite may impede the reverse M → A trans-formation giving rise to stabilization effects on martensite. Morerecently, the transformation behavior of severely deformed NiTiSMAs has been studied [11]. It is observed that heavily deformedB2 and B19′ accompanied with martensite stabilization co-existdue to the formation of the twin-related nanograins by severeplastic deformation. However, the experimental work conductedso far seldom analyzed the transformation behavior of NiTi SMAsdeformed under high strain rates. In some cases of application, e.g.nanocrystallization [16], NiTi SMAs are subjected to severe plasticdeformation at high strain rate. If we are able to predict the transfor-mation points and phases constituent, we will have a good chanceto predict the subsequent shape recovery process accurately. Thus,to understand the transformation behavior of NiTi SMAs deformedat high strain rates is a pre-requisite.

The surface mechanical attrition treatment (SMAT) is an effec-tive technique to generate severe plastic deformation in surfaceregion of the treated sample at high strain rate [17]. This techniquehas been used to enhance the mechanical properties and gener-ate nanocrystalline structures in metals [18–20]. In SMAT, a large

number of metallic balls with 1–10 mm in diameter are resonatedultrasonically. The vibrating balls impact the sample surface fromrandom directions over a short time. Each impact may result inplastic deformation with high strain rates estimated to be as highas 102–103 s−1 in the top surface layer and decreasing to zero in the

nd Compounds 482 (2009) 298–301 299

btawTN

2

wwracDbtoiso3t4b

Cicb2mDmlrTt4swtmu

3

us

Ff

T. Hu et al. / Journal of Alloys a

ulk. Consequently, the repeated multidirectional impacts give riseo severe plastic deformation with gradient variation in the stressnd strain from the surface to the substrate. In the present work,e report for the first time the deformation of NiTi SMAs by SMAT.

he phase constituents and transformation behavior of the treatediTi SMAs are investigated.

. Experimental

A Ti–50.7 at.% Ni plate with a thickness of 3 mm was used in the presentork. Samples were hot rolled and annealed at 973 K for 30 min, followed byater quenching to room temperature. The transformation temperatures of the as-

eceived specimens were determined by DSC to be As = 339 K, Af = 360 K, Ms = 329 Knd Mf = 309 K, respectively. Prior to SMAT, the surface was polished with siliconarbide papers. The detailed SMAT setup and procedures can be found in Ref. [17].uring SMAT, the stainless steel balls were placed in a cylinder chamber agitatedy an ultrasonic generator. When the balls were resonated, the entire surface ofhe sample was impacted by a large number of flying balls within a short periodf time. Each impact of the ball onto the surface resulted in plastic deformationn the surface layer. As a consequence, repeated multidirectional impacts at hightrain rates onto the surface led to severe plastic deformation in the surface layer. Inur experiments, a system vibration frequency of 20 kHz and stainless steel balls ofmm in diameter were selected. The NiTi specimens were treated at room tempera-

ure for 5 min, 15 min, or 30 min. The SMAT processed NiTi plates had dimensions of0 mm × 40 mm × 3 mm and then cut into small blocks about 8 mm × 8 mm × 3 mmy spark discharge for further characterization.

X-ray diffraction (XRD) was carried out on a Philips X’perts diffractometer withu K� radiation (� = 0.154056 nm) at room temperature. By using repeated mechan-

cal polishing, the sample surface was removed layer-by-layer, so that the phaseonstituents at different depths could be determined. The phase transformationehavior was studied by differential scanning calorimetry (DSC) using a TA MDSC910 with a heating and cooling rate of 5 K/min from 213 K to 423 K. The speci-en for DSC analysis was the NiTi sample after undergoing SMAT for 30 min. In theSC measurement, three surface layers at different depths, namely (i) the outer-ost layer (approximate top 150 �m region) denoted as Layer I, (ii) the near surface

ayer (approximate top 150–300 �m region) denoted as Layer II, and (iii) the inte-ior layer (approximate top 400–600 �m region) denoted as Layer III, were probed.o obtain Layer II and Layer III, the surface layer was first removed by grinding andhen mechanically polished from the other side to obtain a 150–300 �m region and00–600 �m region, respectively. With regard to Layer I on the top surface, only theide opposite to the surface of the treated specimen was ground until the thicknessas about 150 �m. The heat of transformation was calculated from the area under

he DSC peaks using the equipment software. Specimens for both the XRD and DSCeasurements were polished manually and prepared below 298 K in order to avoid

ndesired deformation and phase transformation.

. Results and discussion

Fig. 1 shows the surface XRD spectra of the NiTi specimens afterndergoing SMAT for different periods of time. The typical XRDpectrum of the as-received NiTi specimen shows the main phase

ig. 1. XRD patterns acquired from the Ti(50.7 at.%)Ni alloy after undergoing SMATor different durations.

Fig. 2. Variations in the phase structure with depth in NiTi specimen after 30 minSMAT.

of B19′ in addition to a small amount of precipitates identified asTi2Ni, Ni3Ti and Ni4Ti3 phases. After SMAT for 5 min, the inten-sity of the B19′ peak diminishes, whereas the B2 (1 1 0) diffractionpeak corresponding to austenite emerges obviously. With furtherdeformation by SMAT, the B2 (1 1 0) and B2 (2 1 1) diffraction peaksbecome more evident whereas the diffraction peaks of B19′ and pre-cipitates disappear. It is evident that SMAT results in the formationof a parent B2 phase from martensite B19′ in the top layer.

To determine the microstructure evolution with depth, the30 min SMATed NiTi specimen is analyzed by XRD and the XRDpatterns derived at different depths are depicted in Fig. 2. It can beobserved that the treated surface is predominantly composed of theB2 phase. At a depth of about 10 �m, there are obvious changes inthe phase constituents. The B2 intensity diminishes while the B19′

diffraction peaks emerge with intensities comparable to that of theB2 phase. From 20 �m to 40 �m, the intensities of the B19′ peaksincrease. At a depth of about 50 �m, there are noticeable changesin the phase constituents showing a mixture of B19′ and B2 phaseswith B19′ being the main phase. On the other hand, a graded phasestructure is observed to develop from the treated surface extend-ing to the bulk. One may argue that removing the surface region bymechanical polishing may lead to peak broadening or shift in theXRD pattern due to the unavoidable residual strain after mechan-ical polishing. Nevertheless, the small broadening or shift can beneglected compared to the emergence and disappearance of theB19′ and B2 diffraction peaks in the XRD spectra.

Fig. 3 shows the DSC curves of transformation behavior inthe different surface layers after SMAT. In the first heating andcooling cycle, as shown in Fig. 3(a), the critical temperatures ofboth the reverse M → A and forward A → M transformations areunchanged in Layer III (400–600 �m region) compared to the as-received materials. However, in Layer I (0–150 �m region) and LayerII (150–300 �m region), the temperatures As and Af increase bynearly 20 K in the first reverse M → A transformation. The criticaltemperature of the forward A → M transformation shifts to a lowertemperature by 16 K in Layer I and Layer II, whereas the exothermicpeak broadens probably due to small amount of austenite to R phasetransformation [14]. Fig. 3(b) shows the transformation behavior ofthe different surface layers in the second heating and cooling cycle.The shape of the endothermic and exothermic peaks in all the DSC

curves is unchanged in comparison with that in Fig. 3(a). However,the temperatures As and Af of Layer I (0–150 �m region) and LayerII (150–300 �m region) for the second reverse M → A transforma-tion revert to 336 K and 351 K, respectively, which are almost thesame as before SMAT. With regard to the forward A → M transfor-

300 T. Hu et al. / Journal of Alloys and Com

FSa

mctraiis

Dtptch

TTf

LLLA

ig. 3. DSC curves obtained from the as-received sample and top three layers in theMATed NiTi specimen: (a) first heating and cooling cycle and (b) second heatingnd cooling cycle.

ation in the second cycle, little difference among the DSC curvesan be observed compared to the first cycle. The increase in theransformation points in the first reverse M → A transformation andestoration to the original values observed by DSC measurementre known as the martensite stabilization effect [5,8,13,14]. Theres thus evidence that SMAT induces not only the parent B2 phasen the top surface, but also the martensite stabilization effect in theub-surface layer in the NiTi sample.

Table 1 summarizes the transformation heat derived from theSC curves of the SMAT samples in the first and second cycles. The

ransformation heat Q is equal to the area of the corresponding

eak in the DSC curve. Q1

h and Q2h are the transformation heats for

he reverse M → A transformation in the first and second heatingycles, respectively and Q1

c and Q2c represent the transformation

eats of the forward A → M transformation in the first and second

able 1ransformation heat in the first and second heating and cooling cycles determinedrom the SMATed NiTi specimen.

Q1h (J/g) Q1

c (J/g) Q2h (J/g) Q2

c (J/g)

ayer I 14.99 16.15 16.56 18.12ayer II 17.99 19.29 19.05 19.94ayer III 25.70 26.62 24.51 24.61s-received NiTi 25.27 24.96 – –

pounds 482 (2009) 298–301

heating cycles. It is found that the values of Q1h, Q1

c , Q2h and Q2

c ofboth Layer I and Layer II are lower than those of Layer III. Withregard to Layer III, all transformation values are close to those ofthe as-received materials indicating that SMAT has little effects onthe transformation behavior in Layer III which is deeper than LayerI and Layer II.

As shown in Fig. 1, SMAT leads to the formation of a parent B2phase in the surface from the original B19′ phase. The formation ofthe parent B2 phase is similar to Lin’s results [8] in which the B2phase could be induced from the B19′ phase by cold-rolling in theTi50Ni50 alloy. They suggest that the B2 phase induced by stress frommartensite in the Ti50Ni50 alloy is due to the occurrence of the lat-tice softening behavior around the As temperature in cold-rolling.It is suggested that lattice softening may occur in the martensitephase prior to the reverse M → A transformation [21]. In compari-son with cold-rolling, the strain rate at the sample surface duringSMAT is much higher and estimating to be as high as 102–103 s−1.At such a high strain rate, a significant portion of the deforma-tion work is possibly converted to heat due to multiple interfaceinteractions [22]. As a result, the local temperature increases overthe As and Af points and the transformation M → A subsequentlyoccurs. Simultaneously, the lattice softening phenomenon occursand promotes the formation of the stress-induced B2 phase. Webelieve that both lattice softening and deformation heating maycontribute to the formation of the parent B2 phase from martensiteafter SMAT. Moreover, the longer the SMAT time, the more exten-sively the martensite plates are transformed into the parent B2phase. In addition, SMAT results in severe plastic deformation andhigh density dislocations in the entire surface layer. The disloca-tions are believed to hinder the movement of phase interface andso it is difficult for the B2 phase formed during SMAT to transforminto B19′ via the deformation of stress-induced martensite trans-formation (SIMT). That is the possible reason we have not observedthe B19′ phase via SIMT on the top surface of the NiTi specimen afterSMAT. It is also observed that precipitates such as Ti2Ni, Ni3Ti andNi4Ti3 disappear after the SMAT process. The underlying reason isunclear and more work needs to be done.

According to the XRD patterns in Fig. 2, the B19′ phase appearswhile B2 diminishes with depths, indicating that SMAT produces agradient phase structure along from the surface to the bulk. On theother hand, the localized strain created by SMAT decreases gradu-ally from the treated surface to the bulk and hence, it is believedthat formation of the graded phase structure is due to the gradientvariation of strain resulting from SMAT.

The DSC results indicate that the martensite stabilization effectoccurs in the sub-surface layer up to 300 �m of the SMATed NiTispecimen, whereas the stabilization effect cannot be observed inthe bulk. This is due to the gradient variation of strain with depthsafter SMAT. Deformation of martensite usually leads to marten-site variants through martensite reorientation in the NiTi alloy.According to Liu’s theory [14], martensite reorientation not onlycan relax the internal elastic energy which is stored in thermalmartensite, but also can create an internal elastic energy in themartensite variants that oppose its reverse M → A transforma-tion. Moreover, an irreversible energy component is unavoidableduring deformation due to the necessity for plastic deformationas an accommodation mechanism between neighboring grains ofmis-matched variant orientations [23]. Hence, restoration of trans-formation is hindered causing the stabilization effect on martensite.In addition, large numbers of dislocations are introduced due tosevere plastic deformation caused by SMAT. The dislocations in the

deformed martensite structures inhibit the reverse M → A trans-formation by imposing a friction stress on the martensite/parentinterfaces [13]. Upon heating, extra resistance must be overcome inthe reverse transformation, resulting in an increase in the criticaltemperature for the reverse martensite transformation.

nd Com

tssaowtc

Q

wawcsih

4

mBisn

[

[

[[[[

[

T. Hu et al. / Journal of Alloys a

The transformation heat calculated from Layer I and Layer II inhe first heating and cooling cycle is less than that of Layer III, ashown in Table 1. This is due to the reduced transformable marten-ite volume. As indicated by the XRD results in Fig. 2, the B2 phaseppears in the surface layer after SMAT. In other words, formationf the B2 phase in the surface layer lowers the martensite fraction,hich participates in the M → A transformation. In principle, the

ransformation heat of a thermoelastic martensite transformationan be expressed as [14]:

=∫

[�H + Eel(fm) + Eir,w(fm)] dfm (1)

here Eel(fm) is the elastic energy, Eir,w(fm) is the irreversible work,nd fm is the fraction of the martensite. The transformation enthalpyhich is determined by the chemistry and crystallography is a

onstant for a given transformation. The actual amount of marten-ite that participates in the M → A transformation is relatively lown Layer I and Layer II, thereby causing the lower transformationeat.

. Conclusions

SMAT results in the formation of the parent B2 phase from

artensite B19′ in the surface layer of NiTi alloy. The amount of

2 phase decreases with depth and a graded phase structure layers observed. SMAT also induces the stabilization effect on marten-ite in the sub-surface layer up to 300 �m deep in the NiTi alloy buto martensite stabilization effect can be observed in the bulk.

[[[

[[

pounds 482 (2009) 298–301 301

Acknowledgement

Financial support by Hong Kong Research Grants Council (RGC)General Research Funds (GRF) No. CityU 112307 is acknowledged.

References

[1] K. Otsuka, C.M. Wayman, Shape Memory Materials, Cambridge University Press,Cambridge, 1998.

[2] S. Shabalovskaya, J. Anderegg, J. Van Humbeeck, Acta Biomater. 4 (2008)447–467.

[3] K. Wada, Y. Liu, J. Alloys Compd. 449 (2008) 125–128.[4] M. Dolce, D. Cardone, Int. J. Mech. Sci. 47 (2005) 1693–1717.[5] G. Tan, Y.N. Liu, Intermetallics 12 (2004) 373–381.[6] G.N. Dayananda, M.S. Rao, Mater. Sci. Eng. A 486 (2008) 96–103.[7] L.J. Chiang, C.H. Li, Y.F. Hsu, W.H. Wang, J. Alloys Compd. 462 (2008) 47–51.[8] H.C. Lin, S.K. Wu, Metall. Trans. A 24 (1993) 293–299.[9] S. De la Flor, C. Urbina, F. Ferrando, J. Alloys Compd. 469 (2009) 343–349.10] X.M. Wang, B.X. Xu, Z.F. Yue, J. Alloys Compd. 463 (2008) 417–422.

[11] I. Karaman, A.V. Kulkarni, Z.P. Luo, Philos. Mag. 85 (2005) 1729–1745.12] B. Kockar, I. Karaman, J.I. Kim, Y.I. Chumlyakov, J. Sharp, C.-J. Yu, Acta Mater. 56

(2008) 3630–3646.13] H.C. Lin, S.K. Wu, T.S. Chou, H.P. Kao, Acta Metall. Mater. 39 (1991) 2069–2080.14] Y.N. Liu, D. Favier, Acta Mater. 48 (2000) 3489–3499.15] Y.N. Liu, G. Tan, S. Miyazaki, Mater. Sci. Eng. A 438–440 (2006) 612–616.16] T. Waitz, V. Kazykhanov, H.P. Karnthaler, Acta Mater. 52 (2004) 137–147.

[17] K. Lu, J. Lu, Mater. Sci. Eng. A 375–377 (2004) 38–45.18] N.R. Tao, Z.B. Wang, W.P. Tong, M.L. Sui, J. Lu, K. Lu, Acta Mater. 50 (2002)

4603–4616.

19] K. Wang, N.R. Tao, G. Liu, J. Lu, K. Lu, Acta Mater. 54 (2006) 5281–5291.20] H.W. Zhang, Z.K. Hei, G. Liu, J. Lu, K. Lu, Acta Mater. 51 (2003) 1871–1881.21] N. Nakanishi, in: J. Perkins (Ed.), Shape Memory Effect in Alloys, Plenum Press,

New York, NY, 1975, pp. 147–175.22] H.F. Lopez, Mater. Lett. 51 (2001) 144–150.23] Y.N. Liu, Y. Liu, J. Van Humbeeck, Acta Mater. 47 (1998) 199–209.